x = list(x_mean)
x_std = [math.sqrt(v) for v in x_var]
return x, x_std, s
def dlm_set_univariate_params(s, r: R_TYPE):
period_choices = [3, 5, 7, 10, 12, 16, 24, 32]
s['auto_degree'], s['trend_degree'], period_choice = to_int_log_space(
r,
bounds=[
(0.5, 3.2), # auto-reg degree. 1 means linear trend, 2 quadratic
(0.1, 3.4), # seasonality degree
(0, len(period_choices)) # seasonality period
])
s['discount'] = 1 - (
r % 0.05) # Shouldn't matter too much as these get tuned eventually
s['period'] = period_choices[period_choice]
s['n_warm'] = 100
s['n_fit'] = 500 # How often to tune discounts subsequently
return s
if __name__ == '__main__':
err = prior_plot_exogenous(f=dlm_univariate_r3,
k=2,
n=1000,
r=np.random.rand())
plt.figure()
print('done')
pass
from timemachines.skaters.evaluation import evaluate_mean_absolute_error
from timemachines.data.real import hospital_with_exog
y, a = hospital_with_exog(k=k)
y0 = [yi[0] for yi in y]
r = 0.1 # Doesn't matter?
err1 = evaluate_mean_absolute_error(f=f, k=k, y=y0, r=r, n_burn=250)
err2 = evaluate_mean_absolute_error(f=f, k=k, y=y, r=r, n_burn=250)
err3 = evaluate_mean_absolute_error(f=f, k=k, y=y, r=r, a=a, n_burn=250)
errlv = evaluate_mean_absolute_error(f=last_value,
k=k,
y=y,
r=r,
a=a,
n_burn=250)
print('----------------')
print("Error w/o exogenous = " + str(err1))
print("Error w exogenous = " + str(err2))
print("Error w exo + known = " + str(err3))
print("Error last val cache = " + str(errlv))
if __name__ == '__main__':
f = pmd_exogenous
if True:
prior_plot_exogenous(f=f, k=1, n=200, r=0.95)
if True:
prior_plot(f=f, k=1, n=200, r=0.5)
if True:
pmd_exog_compare(f=f)
if y is not None:
# Process observation and return prediction
assert isinstance(y, float) or len(
y) == s['dim'], ' Cannot change dimension of input in flight '
y0, exog = split_exogenous(y=y, dim=s['dim'])
s = update_buffers(s=s, a=a, exog=exog, y0=y0)
if True: # Always fit prior to prediction
none_, s, _ = flux_auto(y=None, s=s, k=k, a=a, t=t, e=e,
r=r) # Fit the model
assert none_ is None
return flux_or_last_value(s=s, k=k, exog=exog, y0=y0)
if y is None:
if len(s.get('buffer')) < s['n_burn']:
s['model'] = None
else:
data = pd.DataFrame(columns=['y'], data=s.get('buffer'))
s['model'] = pf.ARIMA(data=data,
ar=s['ar'],
ma=s['ma'],
target='y',
family=s['family'])
_ = s['model'].fit("MLE")
return None, s, None # Acknowledge that a fit was requested by returning x=None, w=None
if __name__ == '__main__':
err = prior_plot_exogenous(f=flux_auto, k=1, n=200, r=0.05)
pass
return x, x_std, s
def dlm_set_exog_hyperparams(s, r: R_TYPE):
# Univariate model with autoregressive components
# This uses the discounting method of H/W so doesn't need to be fit as often
period_choices = [3, 5, 7, 10, 12, 16, 24, 32]
s['auto_degree'], s['trend_degree'], period_choice = to_int_log_space(
r,
bounds=[
(0.5, 3.2), # auto-reg degree. 1 means linear trend, 2 quadratic
(0.1, 3.4), # seasonality degree
(0, len(period_choices)) # seasonality period
])
s['discount'] = 1 - (
r % 0.05) # Shouldn't matter too much as these get tuned eventually
s['period'] = period_choices[period_choice]
s['n_burn'] = 100
s['n_fit'] = 500 # How often to tune discounts subsequently
return s
if __name__ == '__main__':
err = prior_plot_exogenous(f=dlm_exogenous_r3,
k=1,
n=1000,
r=np.random.rand())
plt.figure()
print('done')
pass